Docker Container variant about anime4k HOT 7 OPEN

Impression was from when I ran everything within docker containers: (I hate to post logs but..)

PS D:\same> docker run -v ${PWD}:/anime 4kanime ruby 4k.rb Aug 15, 2019 11:44:14 AM com.aparapi.internal.opencl.OpenCLLoader <clinit> SEVERE: Check your environment. Failed to load codegen native library or possibly failed to locate opencl native library (opencl.dll/opencl.so). Ensure that OpenCL is in your PATH (windows) or in LD_LIBRARY_PATH (linux). Exception in thread "main" javax.imageio.IIOException: Can't read input file! at java.desktop/javax.imageio.ImageIO.read(ImageIO.java:1308) at com.bloc97.anime4k.Main.main(Main.java:36)

Host OS: Windows 10 Pro
Docker-image: openjdk:12-alpine

Perhaps its an alpine problem, I apologize for not checking if its an alpine problem before reporting, but I shall go do that tomorrow.

For starters, I prefer to not have Java installed on my computer.
Bu main reason being it makes if I were running servers that use the image to post-process image or anything of that sort, deployments will be much simpler than to write scripts to pre-install FFMPEG and JDK & curling the jar binary for every server spun up. Building it into a static image would make my other service dependent on it more readily.

from anime4k.

It works without OpenCL but complains while running.

On Docker Ubuntu apt-get install -y ocl-icd-opencl-dev removes the complaining, but I don't notice a speed difference.

from anime4k.

Oh, you're right about the OpenCL usage. Anime4k apparently uses Arapi which in turn uses OpenCL to execute Java code on the GPU. This means that you need a Docker image with OpenCL support.

from anime4k.

I have created a PR for basic Dockerfile, to build an ubuntu image that's portable.

I also have another one that packs FFMPEG into the image and added a script upscales videos by splitting and merging them (quite slow due to the restart of image kernel each time), should I add a PR for that too?

from anime4k.

Where did you get the impression that it requires OpenCL?

from anime4k.

Regarding Docker container, yes, it could theoretically run inside a Docker container, however, seeing as how it's a realtime algorithm that can easily run on any platform, most likely even the web, I'm not quite seeing the point of using Docker? Is there some use-case that I'm overlooking.

from anime4k.

The algorithm is so simple that it is easy to be bounded by the memory speed. (Copying the image back and forth from the System Memory to the GPU Buffers) If your CPU is powerful enough (Recent CPUs with 8+ cores) it might even outperform GPUs since PCIe lanes are not as fast as the memory bus.

from anime4k.